Abstract: The algorithm that uses SURF algorithm to detect feature points in the process of image matching only uses feature descriptor for image registration with a high probability of mismatch.Usually carried out on the video image stitching,stitching image change is not too big,so is the image corresponding points in the main direction angle and scale factor.This article uses this two characteristics for screening and then take feature descriptor minimum Euclidean distance between points for the correct matching points,achieves multi-feature fusion matching,Increases the matching accuracy,and improves the accuracy of image stitching.